A city permitting cluster needs supported OS images on new GPU nodes. How should the OS be installed?
Select an answer to reveal the explanation.
Short Explanation
Let BCM provision a supported OS onto those permitting nodes—like using the city’s standard image cart instead of random USB sticks. Jobs do not run on bare BMC firmware.
Full Explanation
New GPU nodes in the city permitting cluster must receive a supported OS through BCM provisioning so the image aligns with the municipal stack, driver path, and hardening baseline. BCM imaging keeps OS state consistent under the control plane: the same supported image, same packages, and same identity hooks across the fleet. That is how permitting AI nodes become manageable inventory rather than one-off snowflakes.
Flashing consumer USB sticks with unsupported hobby distros by hand only fails because those media are outside the supported matrix, bypass BCM image control, and create unreproducible nodes that break drivers and compliance. Skipping OS imaging and scheduling Slurm jobs on bare BMC firmware confuses out-of-band management firmware with a host OS—jobs do not run on BMC. Copying OS ISOs onto GPUs as if they were block storage targets misuses accelerators as install media targets and does not provision a bootable, supported host OS via BCM.
Exam caveat: OS install for AI cluster nodes is a BCM provisioning step with a supported image—not hobby USB, not BMC-as-OS, not GPU-as-ISO-disk.
Operational check: assign nodes to the correct BCM category/image, run provisioning, confirm each permitting node boots the expected supported OS, and verify hostname/DNS before driver install.